How the NBA 2027 Desk Prices Spreads & Totals: Fair Lines, Rest Penalties, and Late Scratch Rules
The NBA 2027 Desk prices every 2026–27 regular-season game every 15 minutes until tip-off by combining true bookmaker opening lines, tiered depth-chart starter injury weights (3.5 / 3.0 / 1.5 / 1.0 pts), and non-stacking schedule rest adjustments (B2B -1.0 pt, 3-in-4 -1.5 pts). An official 1.0u Like requires ≥ 2.0 spread points or ≥ 3.0 total points of edge (≥ 4.0 / ≥ 4.5 pts for the 1.5u daily Lock, at most one per slate day, never on a +5 or bigger underdog), capped at one official play per game, skipped when the line has moved 1.0+ pts against our side since the opener, and graded against the closing line.
1. Edge Thresholds & One-Play-Per-Game Rule
Defined in src/lib/nba/edge-thresholds.ts and enforced across the live board and public record. See graded totals · How we bet the first month of the season
| Tier | Unit Stake | Spread Edge Bar | Total Edge Bar | Volume Cap |
|---|---|---|---|---|
| Lock | 1.5u (-110) | ≥ +4.0 pts (SPREAD_LOCK_EDGE) | ≥ +4.5 pts (TOTAL_LOCK_EDGE) | Max 1 Lock per NBA slate day, Pacific date (highest normalized edge among that day's open games; once that day's Lock tips it is final; no Lock on a day with no qualifier). Never on a +5 or bigger underdog. |
| Like | 1.0u (-110) | ≥ +2.0 pts (SPREAD_LEAN_EDGE) | ≥ +3.0 pts (TOTAL_LEAN_EDGE) | Max 1 official play per game (spread or total, never both) |
| Pass | 0.0u | < +2.0 pts | < +3.0 pts | No action — explicit pass reason displayed on game card |
2. Tiered Starter Injury Weights & Freshness Filter
Only confirmed Out depth-chart starters that are fresh for today's matchup move the desk's fair line.
| Starter Tier | Spread Impact | Total Impact | Representative Archetypes |
|---|---|---|---|
| Tier 1 · Superstar | -3.5 pts | -1.5 pts | Nikola Jokić, Shai Gilgeous-Alexander, Luka Dončić, Giannis Antetokounmpo, Jayson Tatum, Victor Wembanyama |
| Tier 2 · All-Star / Primary | -3.0 pts | -1.0 pt | Jaylen Brown, Tyrese Maxey, Paolo Banchero, Cade Cunningham, Karl-Anthony Towns, Chet Holmgren |
| Tier 3 · Solid Starter | -1.5 pts | 0.0 pts | Established two-way rotation starters on the active 5-man depth chart |
| Tier 4 · Low-Usage Starter | -1.0 pt | 0.0 pts | Rookies, spot-start bigs, and low-usage fifth starters |
| Bench / Multi-Week IR | 0.0 pts | 0.0 pts | Non-starters and long-term absences (>7-day return window or reported >24h pre-tip) are already in the opener |
5.5 pts per team; total injury adjustments cap at 3.0 pts per game.isFreshGameInjury): Excludes players whose ESPN status comment references a prior game's weekday, players with a return date more than 7 days after game day, or long-term designations (“re-evaluated in two weeks”, “surgery”, “indefinitely”).3. Schedule Rest Adjustments & Non-Stacking Rule
Computed from the full 2026–27 NBA schedule and prevented from double-counting load-management scratches.
| Schedule Situation | Spread Adjustment | Total Adjustment | Non-Stacking Rule |
|---|---|---|---|
| Back-to-Back (one team, <36h since its last tip) | -1.0 pt on tired team | 0.0 pts | Takes max(1.0, tiredTeamStarterOutPts) — never stacks rest on top of a resting starter |
| 3rd Game in 4 Nights (3-in-4) | -1.5 pts on tired team | 0.0 pts | Same non-stacking rule against the tired team's starter-out points |
| 3-in-4 on a Back-to-Back | -2.0 pts on tired team | 0.0 pts | Same non-stacking rule against the tired team's starter-out points |
| Both Teams Tired (B2B or 3-in-4) | 0.0 pts net spread | -1.0 pt on game total | Total cut applies only beyond any star scoring cut already on the total |
4. True Opening Lines, Market-Move Absorption & Tip-Off Freeze
Each game stores its true bookmaker opening spread (openHomeSpread) and opening total (openTotal). When a game has no fresh injury or rest differential, our fair line equals the market line (0.0 edge) — we never fade steam moves blindly.
When the line has already moved from the opener toward the side our injury or rest read favors, the full move is subtracted from that adjustment, so news the market has priced is never counted twice. When the line has moved 1.0+ pts against our side since the opener, the game is a Pass. Without a stored bookmaker opener, no move is measured.
Lines refresh every 15 minutes until tip-off. At tip-off, the active Lock/Like and closing number are frozen to desk_freezes and graded against the stored close at the logged price (else the stored closing price, else standard -110) on /record.
Which Closing Line Should You Grade CLV Against? Same-Book Close vs. Market Consensus
To measure true Closing Line Value (CLV), grade bets against a market-wide close, not one soft retail book's close. Our desk freezes every NBA play at tip-off and grades it against the Vegas consensus close captured in the final 15 minutes before tip, at the logged or stored closing price (/record). 21+ only. No pick is ever guaranteed.
| Benchmark Type | How It Works | Pros | Cons / Manipulation Risk | How Our Desk Uses It |
|---|---|---|---|---|
Same-Book Retail Close e.g., grading a DraftKings bet only vs. DraftKings' close | Compares your ticket strictly against the final number at the exact retail book where you placed it. | Shows whether you timed that specific book's line move well. | Soft books shade lines to balance recreational liability or lag sharp injury moves, creating “phantom CLV” when the wider market didn't move. | Not used for official record grading. |
Single Sharp Book No-Vig Close e.g., Pinnacle or Circa, vig removed | Uses the highest-limit market-making book's final line and removes the vig to find the true closing win probability. | Highest price-discovery efficiency on sides and totals. | Can briefly skew on isolated limit tests right before tip-off. | Not stored or used for grading. Supported in /clv-calculator through the Close Odds (Your Side) / (Opp Side) no-vig inputs. |
Vegas Market Consensus Close our official grading standard | The pre-tip read (inside the final 15 minutes before tip-off) of the VSIN Vegas consensus spread and total. When no consensus number is posted, a single major book (DraftKings first) or ESPN's line; when no pre-tip read landed, ESPN's posted close. | Avoids single-book outliers and stops anyone from shopping a stale rogue line to manufacture wins. | The fallback can be one book's number, and a consensus price isn't always posted, so payouts use the logged price, else the stored closing price, else standard -110. | Official standard on /record: frozen in PostgreSQL at tip-off and never overwritten; audited via Veri.bet, TrustMyRecord, War Room Picks, and /api/pickauditor-feed. |
Closes captured before Oct 30, 2026 came from the DraftKings line via ESPN.
Frequently Asked Questions · NBA Pricing Model
How many points is an NBA back-to-back worth against the spread and total?
In our 2026–27 NBA pricing model, a team on the second night of a back-to-back against a rested opponent takes a 1.0-point spread penalty, a 3-games-in-4-nights spot costs 1.5 points, and a 3-in-4 that lands on a back-to-back costs 2.0 points. When both teams are tired, the spread is unchanged and the game total comes down 1.0 point. Rest penalties never stack on top of the tired team's starter-out adjustment (the model takes whichever is larger), and the total cut never stacks on a star scoring cut.
How much do NBA starter injuries move the fair spread and total?
Confirmed Out starters are weighted by four impact tiers: Tier 1 Superstars move the spread 3.5 points (-1.5 points on the total), Tier 2 All-Stars/Primary Options move the spread 3.0 points (-1.0 point on the total), Tier 3 Solid Starters move the spread 1.5 points, and Tier 4 Low-Usage Starters move the spread 1.0 point (capped at 5.5 spread points per team and 3.0 total points per game). Bench absences and multi-week injured-reserve absences already priced into the opening number are excluded.
What edge threshold is required for an official NBA Lock or Like?
A 1.0u Like requires at least a +2.0 point spread edge or +3.0 point total edge between our fair line and the live market number. A 1.5u daily Lock requires at least a +4.0 point spread edge or +4.5 point total edge (capped at one Lock per NBA slate day and one official play per game; a day with no qualifying game has no Lock). A spread underdog getting +5 or more points is never the Lock (it can still be a Like), and no Lock or Like is posted when the line has moved 1.0+ points against our side since the true bookmaker opener.
Why does the model never blindly fade NBA steam moves?
When an NBA spread or total moves from the true bookmaker opening number toward the side our injury or rest read favors, the market has already priced that news: every point of that move is subtracted from our injury and rest adjustment, so only the unpriced remainder can create an edge. When the line has moved 1.0+ point against our side since the opener, we pass. With no unpriced catalyst, our fair line equals the market line and we never fade steam blindly.
Which closing line does your desk use to grade picks and calculate CLV?
Every NBA pick is graded against the closing number our odds feed captures in the final 15 minutes before tip-off: the VSIN Vegas consensus line, or a single major book (DraftKings first) or ESPN's line when no consensus number is posted, and ESPN's posted close if no pre-tip read landed. That close is frozen in PostgreSQL and never overwritten. Results are paid at the price logged with the pick, else the stored closing price for that side, else standard -110. CLV is the published release line versus that close, and shows as not tracked when no release line was recorded. In our /clv-calculator you can also enter any sharp book's two-way closing odds (such as -105 / -115) to remove the vig and calculate true closing win probability and EV%. Closes captured before Oct 30, 2026 came from the DraftKings line via ESPN.